Artificial Intelligence, Financial Adviser

Username999

"Artificial Intelligence Will Replace Your Financial Adviser - And That's A Good Thing"


https://www.forbes.com/sites/jayadkisson/2019/01/23/artificial-intelligence-will-replace-your-financial-adviser-and-thats-a-good-thing/#789412cee6b4


...

Reaper

I haven't read it as it won't let me in with my ad blocker on but I agree in principle AI will eventually replace IFAs.

It seems to me an IFA looks at a customer profile, objectives and attitude to risk and matches those to the options available. It seems a very rule based processed, and needs a lot of knowledge of all the possible options. Both are ideally suited to computer processing and I see no reason why a computer cannot match or even exceed the performance of a top IFA.

Furthermore monitoring can be done continuously rather than in an annual review.

I would be interested to hear what the IFAs on the forums think.

kuratowski

Speaking as a customer of an IFA (not an IFA) I don't believe the technology has yet been developed which can understand the morass of legislation relating to UK taxes and pensions! So I am quite happy to pay for a regulated, insured IFA to help to steer me through it all, and one who is also able to meet me face to face which I don't think a bot will ever do...

Edit: As an aside, I find the more you understand of machine learning and how it works, the less inclined you are to trust it.

fred246

Reaper wrote: »

I would be interested to hear what the IFAs on the forums think.

I think it's pretty obvious what they will say.

Aegis

Reaper wrote: »

I haven't read it as it won't let me in with my ad blocker on but I agree in principle AI will eventually replace IFAs.

It seems to me an IFA looks at a customer profile, objectives and attitude to risk and matches those to the options available. It seems a very rule based processed, and needs a lot of knowledge of all the possible options. Both are ideally suited to computer processing and I see no reason why a computer cannot match or even exceed the performance of a top IFA.

Furthermore monitoring can be done continuously rather than in an annual review.

I would be interested to hear what the IFAs on the forums think.

I think we'll see AI advisers in the not too distant future, with gradually more and more complex tasks being taken on by the AI. I'm not sure what I'll do with myself if they get so powerful that they can take over my whole job, but I'm sure I can retrain if necessary.

MaxiRobriguez

There will be more AI. You only have to look at this forum to see the consistency of responses to people asking guidance on what to do with some spare money. Pension for tax benefits, global index tracker for diversified growth...

The difficulty AI will have is determining confidence in peoples answers, for example accurately gauging who honestly can tolerate drops of 50% at anyone given time and those who say they can, but then act on impulse when time comes. IFAs (ie humans) are emotional beings that are flexible and inquisitive. AI is still code which is based on black and white rules currently.

Lomcevak

kuratowski wrote: »

Speaking as a customer of an IFA (not an IFA) I don't believe the technology has yet been developed which can understand the morass of legislation relating to UK taxes and pensions! So I am quite happy to pay for a regulated, insured IFA to help to steer me through it all, and one who is also able to meet me face to face which I don't think a bot will ever do...

I agree - I think regulation and compliance would be a barrier here. The cost of your bot giving duff pensions transfer advice might give pause for thought.

Edit: As an aside, I find the more you understand of machine learning and how it works, the less inclined you are to trust it.

My work involves AI and ML quite heavily, and I agree with this too. The technology is reasonably good at crunching through large amounts of rote work though, so there's a risk to any profession where the current value-add is little more than mechanically punching numbers into a model. However, advisors that bring more than that - e.g. a knowledge of regulation and legislation, which doesn't translate well into the ML domain - should be safe for a while.

Linton

How do you sue an AI advisor for bad advice? One of the characteristics of current AIs is that it is virtually impossible to determine why it came up with a particular answer as a high level explanation expressed in terms understandable to a subject expert. This would make the job of the regulator somewhat difficult.



One can see how an AI could do some tasks, but not the real role of an IFA in understanding the customer, their objectives, fears, wishes, personal circumstances etc, information that cannot satisfactorily be acquired from a tick in the box questionaire.

SonOf

I don't know about that article but with the failure of another robo adviser this week, the financial press has included comments that there is a realisation creeping in that robo-advice is not going to take over. The main issues being that it costs too much for robo-advice firms to "buy" customers. And most of the customers they are buying are low value young investors with no money. So, the contributions are small and unprofitable.

The problem is that many robo-advice firms are not actually that much cheaper or not cheaper at all than IFAs. Increasingly, IFA platforms are heading into the 0.1x% range for charges and there is software development on the IFA side that will see small IFAs able to offer a platform with them able to set zero platform fees for the client.

It seems to me an IFA looks at a customer profile, objectives and attitude to risk and matches those to the options available.

The problem is that risk profiling by questionnaire solely has been found to be wholly unreliable. The FCA have for over a decade said that questionnaires are insufficient when it comes to risk profiling and often lead to the incorrect outcome. They consider questionnaires as a starting point leading to further discussion and then consideration of capacity for loss and knowledge and behaviour by the individual.

Often what people put on the questionnaires is conflicting. Many questionnaires will ask the same question multiple times but in different ways to check the consistency. And often you get a totally different answer by asking it in a different way. Then you have the people what want all the upside but none of the downside. And the people that cannot afford higher risk or do not need higher risk. An IFA should, through discussion, cover behaviour and understanding and adjust the risk profile to what they think it should be. Robo is nowhere near that ability yet and wont be for some time.

If you ask someone to fill out a form with their personal information on, you can bet it will come back incomplete and inaccurate in a number of places and require further questions and answers and discussion to get what is needed. Robo relies on input and if the input is incomplete or inaccurate then the outcomes will be too.

Compliance has an issue as well. People generally dont read reports and information that is given to them. IFAs are frequently asked to put things into simple terms. A visit will frequently see the person hand over a draw load of papers and ask the IFA to go through them or be handed a report/letter and asked to explain it. During reviews, people are not generally that interested in the investments themselves but more about the overall situation and how things will be for the future and doing what is needed, what about xyz going on in the world etc. The actual investments that the adviser spends a lot of time on are frequently the least things looked at by the client.

Then there is the face to face aspect. The older you are, the more likely you are to want to deal with a real person. Whether that remains the case with young people today is an unknown. Also, many older people are lonely and for local IFAs with an office in the village, you often get people just drop in and have a cuppa. Sometimes we feel more like a cafe! Often people will drop in with a form they have on a non advice issue and ask us to explain it to them. And extremely frequently, we are used as sounding boards for concepts and ideas that the person has. We have people who ask us how much they can spend each year or if they can buy a car or how much they can spend on their holiday. More time is spent by an IFA talking about things and planning ahead than there is about the investments.

It seems a very rule based processed, and needs a lot of knowledge of all the possible options.

There was research done many years ago regarding flowcharting of advice outcomes through questions with a view to simplifying and computerising advice for lower cost. If found that despite best efforts, it failed to get to the best outcome in 1 in 4 cases. And those cases were relatively simple. They didn't try anything like bed & ISA/bed & pension, ISA to pension, CGT sales to use allowances, carry forward etc.

AI could handle the really simple stuff. However, IFAs are not much getting involved in the really simple stuff nowadays unless it is with older clients or people that just don't want to do it at all.

AI will get there one day but it won't kill off advisers as there will always be people that want something different. I suspect adviser numbers will drop but firms will become like butchers and fishmongers and be seen as a lifestyle choice for quality. Plus, the market is not big enough or profitable enough for the number of AI players that are trying to get a foothold. They are more likely to kill off other AI providers than kill off IFAs at the moment.

Reaper

Thanks, that was a well considered and useful post.

SonOf wrote: »

And those cases were relatively simple. They didn't try anything like bed & ISA/bed & pension, ISA to pension, CGT sales to use allowances, carry forward etc.

To be honest I think that's actually the easy part for a computer. Having a single system that knows about every option and is up to date on tax laws would take a while to set up but would then be relatively easy to maintain and keep up to date. Only 1 system would need to be kept up to date, rather than every IFA in the country.

I suspect such a tool may be a starting point and initially used by IFAs themselves. You type in the details and it spits out suggestions. It could deskill the role making the IFAs job essentially human data gathering and computer input.

Eventually it might creep up the process towards the start point but as you say it will be hard to ever completely replace the initial customer interaction and get accurate results.

bostonerimus

IF AI IFAs make it easier for people to abdicate personal control and responsibility over their financial decisions then I am against them. AI depends on the underlying algorithms and if they are dictated by financial professionals I see little advantage for the customer as many will be confused and charged unnecessary fees...and potentially lumbered with silly portfolios.